hn-podcast-transcribe

by terrycarter1985

Install this Skill

curl -fsSL -o hn-podcast-transcribe.zip https://skill.seembox.com/api/v1/skills/hn-podcast-transcribe/download && unzip hn-podcast-transcribe.zip

Download hn-podcast-transcribe.zip

SKILL.md About

---
name: hn-podcast-transcribe
description: "Automatically fetch, download, transcribe, and archive Hacker News podcast episodes. Use when: (1) scanning HN for podcast links, (2) downloading podcast audio from URLs, (3) transcribing audio with local Whisper, (4) setting up automated podcast archival pipelines, (5) generating SRT subtitles from podcast audio. Supports Apple Podcasts, Spotify, SoundCloud, direct MP3 URLs, and any yt-dlp-compatible source."
---

# HN Podcast Transcribe

Automated pipeline: Hacker News → podcast discovery → audio download → Whisper transcription → structured archive.

## Quick Start

### One-shot: scan HN and transcribe all found podcasts

```bash
python3 scripts/pipeline.py --model base --archive ./archive --srt
```

### One-shot: transcribe a specific podcast URL

```bash
python3 scripts/pipeline.py --url "https://podcasts.apple.com/..." --model base --archive ./archive
```

### Step-by-step (more control)

1. **Find episodes** from HN:
   ```bash
   python3 scripts/fetch_episodes.py --limit 50
   ```

2. **Download audio**:
   ```bash
   python3 scripts/download_audio.py "https://..." --outdir ./downloads
   ```

3. **Transcribe**:
   ```bash
   python3 scripts/transcribe.py ./downloads/episode.mp3 --model base --srt
   ```

## Scripts

| Script | Purpose |
|--------|---------|
| `pipeline.py` | Full pipeline (fetch → download → transcribe → archive) |
| `fetch_episodes.py` | Scan HN API for podcast links, or accept direct URL |
| `download_audio.py` | Download audio via yt-dlp |
| `transcribe.py` | Transcribe with local Whisper, output txt/json/srt |

## Key Options

- `--model`: Whisper model (tiny/base/small/medium/large). Default: `base`
- `--archive`: Archive directory. Default: `archive/`
- `--srt`: Generate SRT subtitles alongside plain text
- `--url`: Skip HN scan, process a single URL directly
- `--limit`: Number of HN stories to scan (default 50)

## Requirements

- Python 3.8+
- `openai-whisper` (`pip install openai-whisper`)
- `yt-dlp` (`pip install yt-dlp`)
- ffmpeg (system package)

## Automated Scheduling

See [references/workflow.md](references/workflow.md) for cron-based scheduling and archive structure details.

_meta.json Metadata

{
  "ownerId": "kn72jmd20ws94jr0p8b3p24wyn82abgj",
  "publishedAt": 1778095524952,
  "slug": "hn-podcast-transcribe",
  "version": "1.1.0"
}